Potential Barriers to Healthcare in Malawi for Under-Five Children with Cough and Fever: a National Household Survey

J HEALTH POPUL NUTR 2014 Mar;32(1):68-78 ©INTERNATIONAL CENTRE FOR DIARRHOEAL ISSN 1606-0997 | $ 5.00+0.20 DISEASE RESEARCH, BANGLADESH Potential Barriers to Healthcare in Malawi for Under-five Children with Cough and Fever: A National Household Survey

Marte Ustrup1, Bagrey Ngwira2, Lauren J. Stockman3, Michael Deming3, Peter Nyasulu4, Cameron Bowie2, Kelias Msyamboza2, Dan W. Meyrowitsch1, Nigel A. Cunliffe5, Joseph Bresee3, Thea K. Fischer3

1Department of Public Health, Faculty of Health Sciences, University of Copenhagen, Copenhagen, Denmark; 2Department of Community Health, College of Medicine, University of Malawi, Blantyre, Malawi; 3Centers for Disease Control and Prevention, Atlanta, GA, USA; 4Division of Epidemiology and Biostatistics, School of Public Health, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa; 5Department of Clinical Infection, Microbiology and Immunology, Institute of Infection and Global Health, University of Liverpool, Liverpool, United Kingdom

ABSTRACT

Failure to access healthcare is an important contributor to child mortality in many developing countries. In a national household survey in Malawi, we explored demographic and socioeconomic barriers to healthcare for childhood illnesses and assessed the direct and indirect costs of seeking care. Using a cluster-sample design, we selected 2,697 households and interviewed 1,669 caretakers. The main reason for households not being surveyed was the absence of a primary caretaker in the household. Among 2,077 children aged less than five years, 504 episodes of cough and fever during the previous two weeks were reported. A trained healthcare provider was visited for 48.0% of illness episodes. A multivariate regression model showed that children from the poorest households (p=0.02) and children aged >12 months (p=0.02) were less likely to seek care when ill compared to those living in wealthier households and children of higher age-group respectively. Families from rural households spent more time travelling compared to urban households (68.9 vs 14.1 minutes; p<0.001). In addition, visiting a trained healthcare provider was associated with longer travel time (p<0.001) and higher direct costs (p<0.001) compared to visiting an untrained provider. Thus, several barriers to accessing healthcare in Malawi for childhood illnesses exist. Continued efforts to reduce these barriers are needed to narrow the gap in the health and healthcare equity in Malawi.

Key words: Healthcare surveys; Health expenditure; Health services accessibility; Malaria; Pneumonia; Malawi

INTRODUCTION inequities are not only widespread across countries but also within countries. Children from the poor- Worldwide, an estimated 8.8 million children aged est households are more likely than children from below five years die annually. Acute respiratory in- wealthier households to be exposed to health risks, fection is the primary cause of these deaths, account- to be malnourished, to experience reduced access ing for 18% of under-five mortality, followed by di- to preventive and curative healthcare services and, arrhoea and malaria (1). The vast majority of child consequently, to die in childhood (1-5). deaths occur in developing countries, and health Adequate access to and utilization of healthcare Correspondence and reprint requests: Dr. Marte Ustrup services are crucial to improve child health in de- Department of Public Health veloping countries (2,3). However, the rate of ob- Faculty of Health Sciences taining care from a trained healthcare provider re- University of Copenhagen mains low in many developing countries; instead, Oester Farimagsgade 5 children are often treated at home, by an untrained DK-1014 Copenhagen care provider, or not treated at all (2-4). Studies have Denmark Email: [email protected] demonstrated that multiple barriers to healthcare Fax: +45 35327487 exist. Geographic accessibility of facilities is a key Barriers to healthcare of under-five children Ustrup M et al. determinant of utilization, and factors, such as ru- 20% of children have been respectively 2.2% and ral residency, long distance, and high travel costs, 2.7% during the same period (18). Thus, pro-poor have been shown to reduce accessibility (6-8). Eco- targeting has not yet resulted in equal progress in nomic affordability is another major determinant, mortality reduction across socioeconomic strata. and there is ample evidence that low household Consequently, inequities in health and access to income and high care-seeking costs are barriers healthcare disfavouring the poor have persisted to healthcare (6-10). In addition, a wide range of and widened (5,17,18). This situation may poten- demographic factors have been identified to affect tially affect Malawi’s progress towards the United utilization of services, including age, sex, educa- Nations’ Millennium Development Goal to reduce tional level, ethnicity and religion, socioeconomic child mortality (17). status, and family-size and composition (6-8,10). Finally, cultural attitudes and beliefs of the popula- To take appropriate measures towards improved tion influence their healthcare utilization patterns and equitable access to healthcare services, the de- (6-8). There is a need for analyzing the interrelation terminants of limited access need to be identified between the different restrictive factors in more de- in the specific context of Malawi (2,3,18,19). The tail (6). overall aim of the present study was to assess potential barriers to healthcare in Malawi for children The healthcare delivery system in Malawi is three- aged below five years, with cough and fever. We tiered, consisting of primary, secondary, and tertia- studied the interrelated effects of demographic and ry-care levels. Sixty-eight percent of health services socioeconomic predictors of obtaining care from are provided by the public sector and 32% by the trained providers. Furthermore, we estimated the private sector, including the not-for-profit Chris- direct and indirect costs of seeking care, thereby tian Health Association of Malawi (CHAM). In ad- providing detailed cost data which are sparse in the dition, traditional healers are widely used (11,12). literature (6,9). To ensure effective services and equitable access, the Government of Malawi implemented an Essen- MATERIALS AND METHODS tial Health Package (EHP) in 2002 and launched a Study site and population sector-wide approach in 2004 as the vehicle to de- liver EHP services which are provided free of charge The Republic of Malawi is a small sub-Saharan Afri- (13). However, the healthcare system has been can country bordering Zambia and Tanzania to the constrained in the provision of services in recent north and partly embedded into Mozambique to years due to a number of factors (11). The finan- the south. The total population in 2011 was 15.4 cial resources for health service delivery have been million (20). Malawi is among the poorest and the inadequate and unpredictable. Consequently, Ma- least developed countries in the world, ranking 171 lawi’s EHP services have been underfunded since out of 187 countries in the Human Development its implementation (11,14). Furthermore, there has Index (20). Eighty percent of the population lives been an increasing shortage of medical staff since in rural areas, and approximately 74% lives below 1990 as a result of poor working conditions, low the poverty line and 40% in severe poverty (20). wages, deaths caused by the HIV/AIDS epidemic, The life expectancy at birth is 54.2 years, and the and migration of skilled personnel to developed infant and under-five mortality rates are 53 and countries (11,15). Quality of care has been further 110 per 1,000 livebirths respectively (20). compromised by a periodic stock-out of essential drugs and medical supplies (11). Despite these chal- The transmission of malaria is perennial in Malawi, lenges, Malawi has achieved significant reductions with a peak in the rainy season from November to in the infant and under-five mortality rates, with April (21). The transmission of acute respiratory an annual average decline of 4.3% and 4.7% re- infections peaks in the beginning of the cold dry spectively during the period 1990-2011 (16). Key season in April to June (22). factors responsible for this progress include consis- Study design and data collection tent investment in child survival interventions and strong coordination between the Government of We conducted a national household survey from Malawi and the development partners (17). How- the end of February to mid-April 2005, using a clus- ever, while national averages have improved, the ter-sample design with compact segments (23,24). poorest children have benefited the least from this We aimed for a nationally-representative sample of progress (18). The annual average reduction rates 600 children, equaling 3,000 households. Sample- in infant and under-five mortality for the poorest size calculation took into account the expected pro-

Volume 32 | Number 1 | March 2014 69 Barriers to healthcare of under-five children Ustrup M et al. portion of interest (i.e. the proportion of children Figure 1. Flowchart of selecting participants with cough and fever seeking care from a trained healthcare provider), the level of precision de- sired, and the expected design effect (23). Malawi’s Households selected for survey 1998 Population and Housing Census provided a (N=2,697) complete list of enumeration areas (EAs) (25). The projected number of households in each EA was Households not interviewed calculated based on census data, then divided by (n=87) due to: 100 and rounded to the nearest integer, which was • No eligible respondent at home (n=36) considered to be the size of the EA in segments. • Household members absent (n=5) This predetermined segment-size was based on the • Refusal (n=2) prediction that the survey team could complete in- • Household vacant (n=7) terviews in 100 households per day. Subsequently, • Household not found (n=7) 30 EAs were chosen by systematic sampling with • Other (n=30) probability of selection proportional to the size of each EA in segments (23). When the survey team Households completing interview arrived at an EA, it was divided into the predeter- (n=2,610) mined number of segments, using sketch maps, such that each segment in the EA had approximately the same number of households. Among Households without primary caretaker (n=871) these segments, one segment was then selected at random (23). All households in the segment were visited by interviewers who interviewed all primary Primary caretakers identiﬁed in the caretakers of children aged less than five years; if households (n=1,787) found absent, households were revisited later the same day. Data were entered into Personal Digi- Caretakers excluded for not ® tal Assistants (PDAs), using Visual Basic software meeting inclusion criteria (n=118) (version 6) (Microsoft Corporation, Redmond, WA, USA) and analyzed with the SAS-callable version Caretakers completing interview (n=1,669) of SUDAAN® software (version 9.01) (Research Tri- angle Institute, Research Triangle Park, NC, USA) to take account of the cluster-sampling design. Data Children included in the study (n=2,127 ) collection was performed by one survey team of 6 interviewers and 2 supervisors. Children excluded for not meeting In total, 2,697 households were selected for the sur- inclusion criteria (n=146) due to: vey and visited by survey teams. Figure 1 shows a • Missing data on age (n=50) flowchart of inclusion and exclusion of study partic- • Missing or inconsistent data on illness ipants. Interviews were completed in 2,610 (96.8%) • symptoms (n=96) of the households; reasons for non-completion are listed in Figure 1. Among the surveyed households, Children included in main analysis 871 households had no primary caretaker. Thus, a (n=1,981) total of 1,787 primary caretakers were identified; 1,669 (93.4%) of them were eligible for the survey, completed if the child had cough and fever during i.e. had at least one child below five years of age the two weeks preceding the survey, which was de- and were interviewed. termined by a screening question in the beginning of Part II of the questionnaire. A third part of the The first part of the questionnaire recorded the questionnaire was completed if the child had diar- demographic and socioeconomic characteristics rhoea during the two weeks preceding the survey of the household, including access to safe water, Variables and was completed by all caretakers identified for the survey (26). The second part focused on ill- Healthcare utilization (dependent variable): We iden- ness symptoms and healthcare utilization among tified 12 different types of healthcare providers children aged below five years. This part was only in Malawi and defined healthcare utilization as

70 JHPN Barriers to healthcare of under-five children Ustrup M et al.

Figure 2. Distribution of healthcare providers visited among children below ﬁve years of age, with cough and fever, Malawi 2005

25 Trained healthcare provider

20 Untrained healthcare provider No care sought

15 cent

Per 10

0 † ‡ ¶ HealerOther Pharmacy Street vendor Private hospital No care sought GovernmentPrivate hospital health centre Government health centre* Mobile clinic or ﬁeld worker *Includes government health centres and outpatient clinics; †Includes private health centres and clinics; ‡Includes mobile clinics and health surveillance assistants (HSA); ¶Includes household members, neighbours, church, and shops/market; n=433 primary visits to any of those. Providers were cat- The distribution of the summed up asset scores egorized as being trained or untrained based on was right-skewed, i.e. the bulk of households had whether or not they held a formal medical degree asset scores below the mean value. Consequently, or medical qualifications. Trained providers were if divided into quintiles, the lowest wealth quin- further categorized as working in private or govern- tiles would only slightly differ from one another in ment healthcare facilities (Figure 2). terms of asset scores, and a significant number of households with identical asset scores would group Age of children: The following 5 age categories were into different wealth quintiles due to the heaping used: <12, 12-23, 24-35, 36-47, and 48-59 complet- of scores. Therefore, we opted to divide the sample ed months. In the analyses, age categories were di- into terciles. In the analyses, wealth quintiles were chotomized into <12 months and 12-59 months of dichotomized into the poorest (the two lower ter- age due to statistical uncertainty around the small ciles) and the wealthiest (the upper tercile) due to number of subjects in 5 categories. statistical uncertainty around the small number of subjects in three categories. Education of mother: Mothers were asked if they had ever attended school, with ‘yes/no’ answers. Other demographic variables: Child’s sex (male/female) and place of residence (urban/rural) were also Wealth index: We defined socioeconomic status included. in terms of assets. Data on household assets were combined into a wealth index by principal com- Severity of illness: In our screening question, we ponents analysis. Households were assigned a identified children who had cough and fever dur- standardized score for each asset, such as drink- ing the two weeks preceding the survey. Additional ing-water source, type of toilet facility, electricity, symptoms, reported by caretakers, were collected and vehicles, according to a Malawi-specific index only for children who had both of these symp- (27). The scores were summed for each household. toms. Malawi’s Ministry of Health has adopted the Households were ranked and divided into terciles. guidelines of the World Health Organization’s In-

Volume 32 | Number 1 | March 2014 71 Barriers to healthcare of under-five children Ustrup M et al. tegrated Management of Childhood Illness (IMCI) Statistical analyses for the case management of childhood illnesses at the first-line health facilities (28). We applied the We used univariate analyses to assess the effect of IMCI classifications, with modifications, to catego- the six predictive variables (child’s age, sex, educa- rize the symptoms of illness. We defined two illness tion of mother, wealth index, place of residence, categories each comprising two subcategories: mod- and illness severity) on seeking care from a trained erately-severe illness comprising “malaria without healthcare provider compared to seeking care from pneumonia” and “malaria with pneumonia”; and an untrained provider or not seeking care at all. severe illness comprising “severe pneumonia or Only primary healthcare visits were included and, very severe febrile disease”, and “possible serious thus, successive visits were excluded from analysis. bacterial infection.” Children aged 2 months up to Odds ratios (ORs) and 95% confidence intervals 5 years with cough and fever without any danger (95% CI) were calculated for each variable, and the signs were classified as having “malaria without p value of a Wald F-test was evaluated to determine pneumonia”, and children with cough and fever whether associations were significant (p<0.05). accompanied with fast breathing but without any All six variables and all two-factor interaction danger signs were classified as having “malaria with terms were entered into a multiple logistic regres- pneumonia.” Children aged 2 months up to 5 years, sion model. This model was reduced to the final with cough and fever and any of the following dan- model by means of a hierarchical backward elimi- ger signs: being unable to eat or drink, convulsions, nation procedure. Significance was determined by lethargy, unconsciousness, and/or stiff neck, were evaluating the p value of Wald F-test (p<0.05). If classified as having “severe pneumonia or very se- an independent variable was not significant but, vere febrile disease.” Children aged 1 week up to when removed, it changed the OR of a significant 2 months, with cough and fever, were classified as variable by ≥10%, and it was retained in the model having “possible serious bacterial infection.” to adjust for confounding effect; t-tests were used Direct and indirect costs: Costs associated with seek- for comparing means in travel time and direct costs ing care were measured in both monetary terms, i.e. incurred when seeking care with regard to wealth direct costs, and in terms of time, i.e. indirect costs. index, place of residence, and type of healthcare The direct costs included travel costs, user fees for provider visited. Furthermore, we calculated the to- consultation and/or hospitalization, and the costs tal cost each caretaker incurred when seeking care of drugs. Direct costs were measured as the sum and investigated total cost relative to income as de- of out-of-pocket payments in local currency—the scribed elsewhere (31). Malawi Kwacha (MWK). The official exchange rate RESULTS in 2012 was 1 US$=269.5 MWK (29). Indirect costs were measured as time spent travelling and wait- Caretakers provided information on 2,127 children ing when seeking care. The opportunity cost, i.e. (Table 1). The sample of children obtained was the costs of activities forgone due to time spent on representative of the child population in Malawi, seeking care, was assumed to be lost income. Hence, according to the 2004 Malawi Demographic and time spent on seeking care was transformed into Health Survey, with regard to sex and place of resi- lost income per minute as its monetary equivalent, dence (data not shown) (32). A total of 50 children using income data from the Malawi Integrated were excluded from further analysis due to missing Household Survey (30). Waiting time was assumed data on age (Figure 1). Among the remaining 2,077 to be 75 minutes for all caretakers, in accordance children, 504 (24.3%) experienced one episode of with other studies (31). cough and fever in the two weeks preceding the Ethical approval survey, and 28 (1.3%) experienced two episodes, resulting in a total of 532 illness episodes. Illness The study received ethical approval from the Ethics could not be classified in 96 children due to in- Review Committees of the College of Medicine, Ma- complete or inconsistent questionnaire, and these lawi and the Centers for Disease Control and Preven- children were excluded from subsequent analyses. tion, Atlanta, GA, USA, prior to the enrollment of In total, 1,981 (93.1%) children were included in participants. Informed consent was obtained by the the main analyses and, among these children, 433 research assistants for each participant prior to his/ episodes of cough and fever (in 408 children) had her enrollment in the study. Informed consent for occurred in the two weeks preceding the survey. this study conformed to the 45 Code of Federal Reg- Of these episodes, 58.0% were classified as moder- ulations, Part 46.101c and 46.102d requirements. ately-severe illness (47.8% with “malaria without

72 JHPN Barriers to healthcare of under-five children Ustrup M et al.

Table 1. Baseline characteristics of surveyed un- seeking care at all (Table 2). Children aged 12-59 der-five children, Malawi 2005 months (OR 0.52; 95% CI 0.30-0.90) and children living in the poorest households (OR 0.51; 95% CI Household survey 0.30-0.87) were less likely to be taken to a trained Variable (N=2,127) healthcare provider when ill compared to older No. (%) 95% CI children and those living in wealthier households Child’s age (com- respectively. Children with moderately-severe ill- pleted months) ness (OR 0.72; 95% CI 0.47-1.12) were less likely <12 440 (21.2) 19.2-23.4 than children with severe illness to visit a trained 12-23 429 (20.7) 18.9-22.5 healthcare provider; however, this association was 24-35 420 (20.2) 18.3-22.3 not statistically significant. 36-47 316 (15.2) 14.0-16.6 48-59 472 (22.7) 20.8-24.8 In the multiple logistic regression model, child’s Child’s sex age and wealth index continued being associated Male 1,043 (50.1) 48.0-52.1 with seeking care from a trained healthcare pro- Female 1,040 (49.9) 47.9-52.0 vider (Table 2). The effect of potential confounding Education of variables was assessed, and no confounders were mother found. Similarly, no significant interactions were No 654 (43.9) 36.8-51.3 observed. Yes 835 (56.1) 48.7-63.2 Direct and indirect costs Wealth index Poorest 1,340 (68.9) 58.3-77.7 Results of t-tests demonstrated that time spent trav- Wealthiest 606 (31.1) 22.3-41.7 elling when seeking care was significantly associ- Place of residence ated with place of residence and type of healthcare Urban 200 (9.4) 2.6-28.5 provider visited (Table 3). Children living in rural Rural 1,927 (90.6) 71.5-97.4 areas spent more time travelling than those in ur- CI=Confidence interval ban areas (mean: 68.9 minutes vs 14.1 minutes) when visiting any type of healthcare provider. In pneumonia” and 10.2% with “malaria with pneu- addition, children who visited a trained healthcare monia”), and 42.0% were classified as having se- provider spent more time travelling than those vis- vere illness (39.7% with “severe pneumonia or very iting an untrained provider (mean: 76.8 minutes vs severe febrile disease” and 2.3% with “possible seri- 26.2 minutes). ous bacterial infection”). Furthermore, living in the wealthiest households, Healthcare utilization patterns living in an urban area, and obtaining care from a trained healthcare provider and/or from private A trained healthcare provider was visited for 48.0% healthcare facilities were significantly associated (n=208; 95% CI 41.0-55.1) of the illness episodes with higher direct costs (Table 3). However, when that had occurred among the surveyed children in estimating total cost relative to income, caretak- the two weeks preceding the survey. An untrained ers living in rural areas spent a higher proportion provider was visited for 37.9% (n=164; 95% CI (3.2%) of their monthly income when seeking care 31.2-45.1) of illness episodes. The remaining 14.1% than caretakers living in urban areas (3.0%). This (n=61; 95% CI 10.7-18.4) of illness episodes did result was not significant (Table 4). not result in a healthcare visit. The distribution of healthcare providers visited is illustrated in Figure To determine the representativeness of children 2. The most common providers worked in private for whom cost data were available, we compared health centres (n=60; 13.9%; 95% CI 10.0-18.8), those with children for whom data were missing were street vendors (n=60; 13.9%; 95% CI 10.0- by using chi-square tests. Our results indicated that 18.8), or worked in pharmacies (n=49; 11.3%; 95% a higher proportion of children for whom data on CI 8.3-15.2). travel time were available was living in the wealthiest households (52.5%; p=0.04) and was taken to a In univariate analyses, child’s age and wealth in- trained healthcare provider (65.4%; p<0.001) com- dex of the household were associated with seeking pared to children who had missing data on travel care from a trained healthcare provider compared time (47.5% and 34.6% respectively). Furthermore, to seeking care from an untrained provider or not a higher proportion of children who had data

Volume 32 | Number 1 | March 2014 73 Barriers to healthcare of under-five children Ustrup M et al.

Table 2. Univariate and multiple logistic regression analyses for seeking care from a trained healthcare provider among under-five children with cough and fever, Malawi 2005 Univariate Multiple logistic Total No. (%) analyses regression Variable (n=433)* (n=208)† Crude OR p Adjusted OR p (95% CI) value (95% CI) value Child’s age (months) <12 (ref) 85 52 (61.1) 1.00 1.00 12-59 348 156 (44.8) 0.52 (0.30-0.90) 0.02 0.52 (0.30-0.90) 0.02 Child’s sex Male (ref) 197 97 (49.2) 1.00 Female 234 111 (47.4) 0.93 (0.55-1.57) 0.78 NS Education of mother No 194 92 (47.4) 0.99 (0.58-1.69) Yes (ref) 235 112 (47.7) 1.00 0.97 NS Wealth index Poorest 255 112 (43.9) 0.53 (0.32-0.88) 0.51 (0.30-0.87) Wealthiest (ref) 141 84 (59.6) 1.00 0.02 1.00 0.02 Place of residence Urban (ref) 41 23 (56.1) 1.00 Rural 392 185 (47.2) 0.70 (0.27-1.78) 0.44 NS Severity of illness Moderate 251 112 (44.6) 0.72 (0.47-1.12) 0.14 NS Severe 182 96 (52.7) 1.00 *All illness episodes in the two weeks preceding the survey; †Illness episodes for which care was sought from a trained healthcare provider; CI=Confidence interval; NS=Not significant; OR=Odds ratio; Ref=Reference group

Table 3. Results of t-test of travel time and direct costs associated with seeking care for under-five children with cough and fever, Malawi 2005 Travel time (min) Direct costs (MWK)† Variable (n=187)* (n=214)‡ No. Mean (range) p value No. Mean (range) p value Wealth index Poorest 107 72.1 (1.0-660.0) 0.08 137 41.6 (0.0-640.0) 0.01 Wealthiest 74 47.9 (1.0-360.0) 70 117.5 (0.0-900.0) Place of residence Urban 20 14.1 (1.0-60.0) <0.001 22 177.6 (0.0-600.0) 0.02 Rural 167 68.9 (1.0-660.0) 192 54.2 (0.0-640.0) Type of provider Untrained 51 26.2 (1.0-120.0) <0.001 114 34.1 (0.0-500.0) <0.001 Trained 136 76.8 (1.0-660.0) 100 104.4 (0.0-900.0) Type of facility Government 63 59.4 (1.0-360.0) 0.09 48 19.7 (0.0-240.0) <0.001 Private 73 91.8 (1.0-660.0) 52 182.5 (0.0-900.0) *Illness episodes for which information on travel time were available; †Direct costs include travel costs, user fees, and costs of drugs measured in local currency—the Malawi Kwacha; ‡Illness episodes for which information on direct costs were available; Min=Minutes; MWK=Malawi Kwacha

74 JHPN Barriers to healthcare of under-five children Ustrup M et al.

Table 4. Total costs of seeking care relative to monthly household income for under-five children with cough and fever, Malawi 2005 Monthly Total Total house- Direct Travel Waiting Total % of time time Variable Number hold costs time time costs monthly spent costs income (MWK)† (min) (min)‡ (MWK)** income (min)¶ (MWK)§ (MWK)* Urban 11 10,784 213.2 19.0 75.0 94.0 105.2 318.4 3.0 Rural 115 3,353 57.8 70.5 75.0 145.5 50.9 108.7 3.2 p<0.05 p=0.1 p=0.1 p=0.08 p<0.05 p=0.5 *Data from the Malawi Integrated Household Survey (30) measured in the local currency—the Malawi Kwacha; †Direct costs include travel costs, user fees, and costs of drugs measured in the local currency— the Malawi Kwacha; ‡Waiting time was assumed to be 75 minutes for all caretakers in accordance to other studies (31); ¶Sum of travel time and waiting time; §Average cost per minute (calculated as lost income per minute: monthly household income divided by four 40-hour work weeks) multiplied by total time spent; ** Sum of direct costs and total time costs. MWK=Malawi Kwacha; Min=Minutes on direct costs had an educated mother (56.6%; availability of effective treatment. They may also be p<0.01) whereas a lower proportion was taken to a hindered by long distances to health facilities and trained healthcare provider (48.1%; p<0.001) com- by financial barriers (2,3,18,33). Consequently, in- pared to children who had missing data on direct equities in healthcare are still evident in Malawi. In costs (43.4% and 51.9% respectively). contrast to other studies, we found no association between maternal education and healthcare utili- DISCUSSION zation, possibly because of dichotomous definition Our study identifies several potential barriers to of education in our study (3,6). obtaining healthcare from a trained healthcare A substantial proportion of children was taken to provider in Malawi and thereby refines our under- an untrained healthcare provider, particularly to standing of the gravity of these barriers and how street vendors and retail pharmacy workers. These they interrelate. This knowledge is important for sources of treatment are convenient in terms of policy-makers and programme managers in their proximity and are relatively low-cost (19,34). How- planning of equitable delivery of healthcare service ever, studies have raised doubt about the quality of and child survival interventions in Malawi. drug dispensing, such as the provision of ineffec- Overall, we found that nearly half of the children tive drugs and stock-outs of drugs (19,35-38). with cough and fever were taken to a trained health- Our findings further showed that caretakers from care provider. This rate is considerably higher than rural households spend more time travelling than rates reported previously in Malawi and Tanzania those in urban areas when seeking care. This is sup- (32,33). However, this comparison is not straight- ported by similar studies conducted in Malawi and forward because of different case definitions. In ad- Ghana (6,39). Rural areas are often underserved by dition, we demonstrated significant inequities in health facilities and a lack of transport and poor healthcare utilization. We found that infants were road conditions make it even more difficult to trav- two times more likely than older children to be el long distances (6,39). Travel time could be very taken to a trained healthcare provider, suggesting costly in terms of opportunity costs, primarily lost that caretakers in Malawi may be particularly aware income (31,39). Long and difficult travel may not of the vulnerability of infants with cough and fe- only cause delays in healthcare visits and thereby ver. These findings are similar to those reported in aggravate the illness, it might also hinder children other studies (18,32,33). In addition, our results from receiving care (6,31,39). confirmed what has been noted previously in developing countries that children living in the poor- As expected, the direct costs of seeking care in- est households are less likely to obtain care from creased in absolute terms for the wealthiest and a trained healthcare provider when ill (6,9). Care- urban households. When measuring total cost takers from the poorest households may not be relative to income, caretakers living in rural areas well-informed about danger signs of illness and the compared to urban areas spent a higher percent-

Volume 32 | Number 1 | March 2014 75 Barriers to healthcare of under-five children Ustrup M et al. age of their monthly income when seeking care. interventions to improve healthcare-seeking However, this result was not significant. Studies patterns. Health communication efforts could in Zambia and Burkina Faso have reported even reinforce caretakers’ ability to perceive the sever- more regressive healthcare expenditure than those ity of illness and initiate the appropriate response in our study (9,31). Our estimations of time costs (2,19,38). Finally, it should be emphasized that were based on an average income per year. It has multisectoral strategies to reducing poverty and so- previously been reported that healthcare visits in a cioeconomic inequities are crucial to the success of farmer’s high season are even more costly in terms specific interventions as recognized by the Govern- of lost income (31). Healthcare costs often put a ment of Malawi (2,36). major burden on households and, to cope with Future research on healthcare performance by dif- payments, households adopt strategies, such as the ferent care providers in Malawi would improve our sale of assets and borrowing money. These coping evaluation of the likely benefits or hazards of ob- strategies can be catastrophic and may force house- served utilization patterns. Continued research on holds into debt or poverty or push them into deeper the trends in healthcare utilization will help assess poverty (40,41). Furthermore, the direct costs were the success of the initiatives made by the Govern- considerably higher when care was sought from a ment of Malawi and the development partners. private compared to a government facility. Thus, despite the significant role of the private sector in Limitations service provision in Malawi, this sector puts a major economic burden on households seeking care Our study had a number of limitations. The cross- from private facilities. sectional study design precluded the identification of a causal relationship between predictor and out- Implications come variables. In addition, relying on retrospective self-reports increased the susceptibility to recall bias Socioeconomic inequities in healthcare are still and social desirability bias. A two-week recall pe- evident in Malawi and, therefore, addressing bar- riod was used in minimizing the problem of recall. riers to healthcare is a key priority for improving Our illness classifications were presumptive and, child health and achieving the Millennium Devel- as shown elsewhere, some symptoms are not rec- opment Goal to reduce child mortality (18). One ognized by caretakers, which may have led to mis- of the first responses should be improved access of classification (33,42). It should also be noted that the poorest households to healthcare services. This we only studied a subset of pneumonia and malaria could be accomplished by reducing logistical and episodes according to the IMCI guidelines, namely financial barriers. Geographic availability of health- those with cough and fever. Another limitation was care facilities could be improved through targeted that the study was conducted during the rainy sea- interventions, increased availability of health fa- son in Malawi and, thus, the prevailing illness and cilities in rural areas, and improved infrastructure healthcare utilization patterns could be different (2,6,18,31). This has been recognized by the Gov- from the dry season. Finally, missing of data in the ernment of Malawi, which increased the number of cost analyses was a limitation of our study, and chil- rural health centres from 219 to 258 between 2003 dren for whom cost data were available were not and 2010 (11). Furthermore, basic training of un- representative of all children. Therefore, our cost trained healthcare providers might ensure higher results should be interpreted with caution. standards of care and referral from these sources. In addition, financial access to care from the private Conclusions health sector could be improved by reducing costs. This could be achieved by implementation of a fair We demonstrated that, in a poor African country, user-fee system where the poorest people are guar- like Malawi, significant inequities exist in obtaining anteed access and treatment through a waiver sys- healthcare from a trained provider for childhood tem, health insurance programmes, or payment by illnesses and in the costs associated with seeking installments (36,43). Since 2005, the Government care. Continued efforts to reduce these barriers are of Malawi has sought to reduce financial barriers to needed to narrow the gap in health and healthcare healthcare by introducing service-level agreements equity in Malawi. with the CHAM’s facilities to allow free access to ACKNOWLEDGMENTS maternal and child health services. By 2010, seven- ty-two of 150 private facilities had such free access We are grateful to the field team members who in- (11). Other priorities could be community-based terviewed the participants. We thank Roger I. Glass,

76 JHPN Barriers to healthcare of under-five children Ustrup M et al.

Centers for Disease Control and Prevention, for his 13. Malawi. Ministry of Health. A joint programme of contribution in the planning of our study. work for a health sector wide approach (SWAp) [2004- 2010]. Lilongwe: Department of Planning, Ministry REFERENCES of Health, Republic of Malawi, 2004:1-72. 1. Black RE, Cousens S, Johnson HL, Lawn JE, Rudan I, 14. Bowie C, Mwase T. Assessing the use of an essential Bassani DG et al.; Child Health Epidemiology Refer- health package in a sector wide approach in Malawi. ence Group of WHO and UNICEF. Global, regional, Health Res Policy Syst 2011;9:4. doi: 10.1186/1478- and national causes of child mortality in 2008: a sys- 4505-9-4. tematic analysis. Lancet 2010;375:1969-87. 15. Record R, Mohiddin A. An economic perspective 2. Victora CG, Wagstaff A, Schellenberg JA, Gwatkin D, on Malawi’s medical “brain drain”. Global Health Claeson M, Habicht J-P. Applying an equity lens to 2006;2:12. doi:10.1186/1744-8603-2-12. child health and mortality: more of the same is not 16. United Nations Children’s Fund. Levels and trends in enough. Lancet 2003;362:233-41. child mortality: report 2012. New York, NY: United 3. Mulholland EK, Smith L, Carneiro I, Becher H, Le- Nations Children’s Fund, 2012. 28 p. hmann D. Equity and child-survival strategies. Bull 17. United Nations. Malawi country assessment report. World Health Organ 2008;86:399-407. Lilongwe: United Nations, 2010. 110 p. 4. Stallings RY. Child morbidity and treatment patterns. 18. Zere E, Moeti M, Kirigia J, Mwase T, Kataika E. Equity Calverton, MD: ORC Macro, 2004. 134 p. (DHS com- in health and healthcare in Malawi: analysis of trends. parative reports no. 8). BMC Public Health 2007;7:78. doi: 10.1186/1471-2458 5. National Statistical Office. Malawi demographic and -7-78. health survey 2010. Zomba: National Statistical Of- 19. Chibwana AI, Mathanga DP, Chinkhumba J, Camp- fice, 2011. 578 p. bell CH, Jr. Socio-cultural predictors of health-seeking 6. Buor D. Analysing the primacy of distance in the uti- behaviour for febrile under-five children in Mwan- lization of health services in the Ahafo-Ano South dis- za-Neno district, Malawi. Malar J 2009;8:219. doi: trict, Ghana. Int J Health Plann Manage 2003;18:293- 10.1186/1475-2875-8-219. 311. 20. United Nations Development Programme. Malawi: 7. Rainey JJ, Watkins M, Ryman TK, Sandhu P, Bo A, Human development indicators. New York, NY: Banerjee K. Reasons related to non-vaccination and United Nations Development Programme, 2011. under-vaccination of children in low and middle (http://hdr.undp.org/en/countries/profiles/MWI, ac- income countries: findings from a systematic re- cessed on 1 April 2014). view of the published literature, 1999-2009. Vaccine 2011;29:8215-21. 21. World Health Organization. 2009 annual report. Li- longwe: Malawi Country Office, World Health Orga- 8. Gabrysch S, Campbell OMR. Still too far to walk: nization, 2009. 38 p. literature review of the determinants of delivery service use. BMC Pregnancy Childbirth 2009;9:34. doi: 22. Vaahtera M, Kulmala T, Maleta K, Cullinan T, Salin 10.1186/1471-2393-9-34. M-L, Ashorn P. Epidemiology and predictors of infant morbidity in rural Malawi. Paediatr Perinat Epidemiol 9. Makinen M, Waters H, Rauch M, Almagambetova N, 2000;14:363-71. Bitran R, Gilson L et al. Inequalities in health care use and expenditures: empirical data from eight develop- 23. United Nations Children’s Fund. Designing and se- ing countries and countries in transition. Bull World lecting the sample. In: Multiple indicator cluster sur- Health Organ 2000;78:55-65. veys. Chapter 4. New York, NY: United Nations Chil- dren’s Fund, 2000:4.1-4.35. (http://www.childinfo. 10. Abebe DS, Nielsen VO, Finnvold JE. Regional in- equality and vaccine uptake: a multilevel analysis of org/files/MICS3_Chapter_4_-_Designing_and_Select- the 2007 Welfare Monitoring Survey in Malawi. BMC ing_the_Sample_060219.pdf). Public Health 2012;12:1075. doi: 10.1186/1471-2458- 24. Turner AG, Magnani RJ, Shuaib M. A not quite as 12-1075. quick but much cleaner alternative to the Expanded 11. Malawi. Ministry of Health. Health sector strate- Programme on Immunization (EPI) Cluster Survey gic plan 2011-2016. Draft III. Lilongwe: Ministry of design. Int J Epidemiol 1996;25:198-203. Health, Republic of Malawi, 2011. 91 p. 25. National Statistical Office. 1998 Population and hous- 12. Africa Health Workforce Observatory. Human re- ing census, population projections, 1999 to 2002. Li- sources for health. Country profile: Malawi. Geneva: longwe: National Statistical Office, 1998. 21 p. World Health Organization, 2009. 55 p. 26. Stockman LJ, Fischer TK, Deming M, Ngwira B, Bow-

Volume 32 | Number 1 | March 2014 77 Barriers to healthcare of under-five children Ustrup M et al.

ie C, Cunliffe N et al. Point-of-use water treatment ni J, Schellenberg JA et al. Retail supply of malaria-re- and use among mothers in Malawi. Emerg Infect Dis lated drugs in rural Tanzania: risks and opportunities. 2007;13:1077-80. Trop Med Int Health 2004;9:655-63. 27. Gwatkin DR, Rutstein S, Johnson K, Pande R, Wag- 36. Malawi. Malawi poverty reduction strategy paper. Li- staff A. Malawi: Socio-economic differences in health, longwe: Republic of Malawi, 2002:1-80. nutrition, and population. Washington, DC: World 37. Marsh VM, Mutemi WM, Muturi J, Haaland A, Wat- Bank, 2000. 28 p. kins WM, Otieno G et al. Changing home treatment 28. World Health Organization. Integrated management of childhood fevers by training shop keepers in rural of childhood illnesses chart booklet. Geneva: World Kenya. Trop Med Int Health 1999;4:383-9. Health Organization, 1997. 39 p. 38. Smith F. The quality of private pharmacy services in 29. Exchange Rates UK. Malawi Kwacha exchange rate, low and middle-income countries: a systematic re- 2012. (http://www.exchangerates.org.uk/Malawi- view. Pharm World Sci 2009;31:351-61. Kwacha-MWK-currency-table.html, accessed on 9 39. Ewing VL, Lalloo DG, Phiri KS, Roca-Feltrer A, Mang- July 2012). ham LJ, SanJoaquin MA. Seasonal and geographic 30. National Statistical Office. Malawi integrated house- differences in treatment-seeking and household cost hold survey 2004-2005. Zomba: National Statistical of febrile illness among children in Malawi. Malar J Office, 2005. 150 p. 2011;10:32. doi: 10.1186/1475-2875-10-32. 31. Hjortsberg CA, Mwikisa CN. Cost of access to health 40. Leive A, Xu K. Coping with out-of-pocket health pay- services in Zambia. Health Policy Plan 2002;17:71-7. ments: empirical evidence from 15 African countries. 32. National Statistical Office. Malawi demographic and Bull World Health Organ 2008;86:849-56. health survey 2004. Zomba: National Statistical Of- 41. Kruk ME, Goldmann E, Galea S. Borrowing and sell- fice, 2005. 454 p. ing to pay for health care in low- and middle-income 33. Schellenberg JA, Victora CG, Mushi A, de Savigny D, countries. Health Aff (Millwood) 2009;28:1056-66. Schellenberg D, Mshinda H et al.; the Tanzania IMCI 42. Hill Z, Kendall C, Arthur P, Kirkwood B, Adjei E. Rec- MCE baseline household survey study group. Inequi- ognizing childhood illnesses and their traditional ex- ties among the very poor: health care for children in planations: exploring options for care-seeking inter- rural southern Tanzania. Lancet 2003;361:561-6. ventions in the context of the IMCI strategy in rural 34. Molyneux CS, Mung’ala-Odera V, Harpham T, Snow Ghana. Trop Med Int Health 2003;8:668-76. RW. Maternal responses to childhood fevers: a com- 43. Su TT, Kouyaté B, Flessa S. Catastrophic household parison of rural and urban residents in coastal Kenya. expenditure for health care in a low income society: Trop Med Int Health 1999;4:836-45. a study from Nouna District, Burkina Faso. Bull World 35. Goodman C, Kachur SP, Abdulla S, Mwageni E, Nyo- Health Organ 2006;84:21-7.

78 JHPN